National Geographic

VOICES Voices Icon Ideas and Insight From Explorers

Menu

Robot Ants Could Make Us More Efficient

With their tiny wires and circuits, robotic ants won’t be taking over the world anytime soon.

But what these artificial insects lack in processing power, they make up for in efficiency: Robotic ants can automatically choose the shortest route from their food sources back to their nests, just like real ants, a new study says. This gives valuable insight into how people should plan transportation and communication systems.

“It’s really interesting to look at social insects because [they] can give us a way to manage information in our societies,” said Guy Theraulaz, a behavioral biologist at the National Center for Scientific Research in France, a co-author on the study. “We take some inspiration from nature.” (Related: “Color-Changing Rubber Robot Could Aid Animal Study.”)

Robo-ants aren’t so different from the insects they mimic. Real ants have tiny brains, which means navigating everyday life, with all its sights and vibrations, is a challenge. So to save brainpower, these insects have evolved to ignore outside stimulation.

“Ants are pretty dumb by themselves,” said study leader Simon Garnier, a biologist at the New Jersey Institute of Technology’s Swarm Lab. “They have about a hundred thousand neurons. There are more neurons in your finger.”

The Path Most Efficient

Despite their simplemindedness, ants almost always take the most efficient path home, which has long stumped the scientific community.

So Garnier and colleagues programmed tiny robots to act like ants using a series of simple computer commands and then put them in a labyrinth. When the robot ants reached a fork in the road, they kept walking straight until they hit an obstacle and veered off in the direction of least resistance—the shortest distance. The ants were relying on simple physics.

“If I blindfold you and put you in the corridor, and you hit the wall, you’re more likely to take the path that deviates less,” said Garnier, whose study was published March 28 in the journal PLOS Computational Biology.

Once the robot ants find the most efficient way through the labyrinth, they alert their peers by calling attention to the pathway with lights. (Real ants use pheromones, or chemical markers.) The robot ants go marching one by one, each laying down a new layer of light. Soon, all ants are traveling on the same, highly productive road. (Watch a video of fierce army ants.)

“It’s like if you want to go to a restaurant with your friends, and one says, ‘I want to go to pizza.’ The other says, ‘I want to go to Chinese food.’ And at one point, if more friends say they want to go to pizza, you’ll go there,” Garnier added.

Combating “Collective Madness”

It’s called the “ant algorithm,” and our societies could benefit from applying it to everyday problems. (Also see “Could Cyborg Cockroaches Save Your Life?”)

Like ants, “we are overloaded with information, but we didn’t develop the appropriate filter,” said Theraulaz, of the French scientific center. “Now we produce a kind of collective madness, and that’s the problem.”

Adopting more collective-swarm intelligence would also make human society less costly and more productive, Theraulaz said. For example, thinking like an ant swarm could better plan shipping routes, place cell phone towers, and task assignments within companies.

The next step is to build more robot-ant studies to further test the algorithm and then apply it to large-scale systems, such as city planning and freeway mapping.

“We can build very simple entities,” Theraulaz said. “This kind of technology will invade society and promote collective intelligence.”

Comments

  1. Yasmine G
    April 15, 2013, 5:24 pm

    Very Insight article!

  2. Tim
    Minneapolis
    April 1, 2013, 3:11 pm

    How could this ant simulation possibly be a more efficient or effective way of visualizing city planning than a computer simulation? What does it offer that a computer simulation cannot? (Not to knock it – it’s a cool project, and I definitely anticipate a slew of other future real world applications)

  3. prapull khanderao
    India
    April 1, 2013, 10:11 am

    Ant Colony optimization technique,. part of AI

  4. ap morgan
    ogun state (nigeria)
    April 1, 2013, 5:37 am

    i have been inspired by your video, i will like to be a part of this huge family where i can share my ideals and my vision.

  5. JDC
    March 30, 2013, 3:41 pm

    Sorry one more point; the ants don’t ‘find’ the most efficient route and then alert their peers. The most efficient routes emerge through the repeated interactions of ants en toto. In other words, tne ant does not make a blueprint which is then followed. Instead, ants follow a few simple rules: Rule number 1: if you find food, pick it up, carry it back to nest while leaving pheromone trail to alert others. Rule number 2: follow the trail back to the food source and repeat Rule number 1 until food is gone. Rule number 3: if you find a fresh trail, follow it to food so you can do Rule number 1 and 2. The result is that other ants come across fresh trails, follow it to food and lay their own pheromone trail down which reinforces the intial route until food is gone. The rules dont simply lead to the most efficient routes, but the ‘single-mindedness’ of ants in following these rules leads to an emergent efficiency. However, to recapitulate, there is never a blueprint that an ant ever knows/develops. Also, it seems worth pointing out that the efficient routes are completely contingent as they emerge and adapt as food sources change. Therefore, a blueprint would become useless every time a food source disappears. Cheers!

  6. JDC
    March 30, 2013, 3:24 pm

    Also, I don’t think it is ‘simple physics’ which guide the interactions or the emergent routes. It is physics, but also chemistry, biology, ecology, etc… Further, calling the ants ‘dumb’ seems to assume a very anthropocentric and limited view of intelligence and cognition in general. Leaving the complex topics of intelligence/cognition aside for now, just think of how complex a single ant is, let alone a colony of ants or the myriad of diverse yet interconnected webs of life which have risen, evolved and fallen for billions of years in order to sustain colonies of ants and humans to observe them…

  7. JDC
    us
    March 30, 2013, 3:18 pm

    Fascinating topic… I would like to encourage some conversation from people who study these types of emergent phenomena. It seems to me that ants don’t ‘automatically’ choose the most efficient routes; rather the ants automatically follow rules for interaction and the most efficient routes emerge between available food sources. Also, emergent properties of complex systems already have ‘invaded’ society; we see them not only in other living species but in our own social environments where a large number of interactions are guided by relatively simple rules. Further, it is not the rules or the algorithm which somehow dictate the end result; instead, complex forms emerge from relatively stable sets of interactions within social and other physical environments, adaptively and contingently. This type of beautiful efficiency/harmony is ubiquitous all around us.

  8. Paul Berry
    Pasadena, CA
    March 30, 2013, 1:07 pm

    Society is not reducing me to the status of an ant fast enough. You are your car.

  9. Charles
    March 30, 2013, 9:36 am

    What is described in the article could be tested using Genetic Programming an a simulated environment. I wonder if the research team has considered this. Algorithms could be discovered much quicker.

  10. jacob beasley
    Minneapolis, MN
    March 30, 2013, 9:12 am

    An Ant Algorithm would just exacerbate traffic jams. Everyone would end up going on the exact same roads to get from point A to point B. Also, people aren’t going to the same destinations. They are going to different ones, so an ant algorithm wouldn’t be appropriate.

  11. KBR
    Spain
    March 30, 2013, 6:50 am

    I for one welcome our new Ant Overlords.

  12. hammeed
    nigeria
    March 30, 2013, 5:06 am

    Well,its qiute fantastict.Anyway i wonder what this could link to interms of development.Perhaps i hold my comment and see how the raminifications pans out

  13. Ima Ryma
    March 30, 2013, 3:48 am

    Humans have been watching how ants
    Manage info with a small brain,
    Simplifying all circumstance
    Into the most efficient chain.
    Not much to do, some humans built
    Robot ants to crawl in a maze,
    Alert in swarm in lit up quilt
    How all should march in the same ways.
    This test somehow is meant to show
    Collective swarm intelligence
    Is maybe how humans should go –
    Time for Star Trek Borg to commence.

    Future humans will get to live,
    Only if we are productive.

  14. Romanovski
    Hawaii
    March 29, 2013, 5:47 pm

    Efficiency is the least of this discovery. Think of outside the box. This type of programming can be applied to AI to be used in robots or computers in learning skills and over time become extremely precise in decision making on any scale. Example: imagine AI cars that have collected data from this type of lab were the car learns every possible way to navigate the roads via the information collected through its sensors and using something to “Simplex Method” to determine best solution on the amount of variables collected. Then apply this data to actual cars on the road with a built in computer with sensors. Not only that but think of GPS data added to the equations. Think of a bigger picture of airplanes being able to make decisions on the fly like we do in different situations. These systems are able to learn from their mistakes and will just become better the more we use them over time. Think of a robot trying to learn how to walk. At first it will fall a lot, but as trial and error creates enough data to correlate the robot will be walking in no time, just like a baby learns how to walk while reaching the stage of toddler.

  15. Chris Kiefer
    San Francisco
    March 29, 2013, 4:38 pm

    One should note that many major American roads were originally the trails of Native Americans and wildlife–including ants! Those paths are already efficient. Could the “ant algorithm” would be more useful if applied to the operation of existing highways, rather than the planning of new ones? I’m thinking of cars communicating with each other during peak traffic.

  16. Mike S.
    March 29, 2013, 4:03 pm

    No matter how many times I click on the video accompanying this story all I get are advertisements – something I absolutely will not watch.

    Why are you hiding your information behind advertisements? You *KNOW* we won’t stoop to being bombarded with more and more advertising – we’ll just figure out a way around it anyway. If you’re going to be on the Internet, why not join the Internet community, and stop trying to bilk us out of our eyeballs all the time?

    Thanks, but no thanks. I’ll just google this story somewhere else…